DolbyA decoding feedback -- 'feral' examples (yes/no)

[John Dyson]

On a happier note -- over these last few weeks, biding time until I can settle down to work more intensively on the C4 -- I have been doing decodes, and more importantly developed some very good generic filters for consumer decoding of material.   There are some common filter patterns that can be selected in segments of LF, MF, HF -- each one is semi independent, and there is a certain pattern that appears to be applicable at least 1/3 of the time.  After that first group of low hanging fruit with the common filter, there are several more combinations that happen fairly often.  My guess -- if I create a selection of 3 versions of LF, 3 versions of MF and 3 versions of HF -- then perhaps 75% of the cases are  handled.   Next, if I include some special filter sets for Linda Ronstadts' recordings that also work in some other cases -- can probably get the percentage up to 90%.  After that, we are far into diminishing returns.

 

I can publish these filter sets along with supporting command line options for the majority of the cases.  If a combination appears to come close, but not hit exactly -- then the user can tweak based upon the built-in filter set values.

 

The goal of the internal filter sets would be to help avoid needing SOX, and the decoder could operate directly on wav files without using command pipeline chains -- which are almost an oddity when used on Windows.  Of course, Windows is where most people 'reside'.  The decoder is built to run more naturally on Unix like OSes, so the pipeline combination concept seems more natural on Unix  (Unix is like a set of toolkits, not so much monolithic do-it-all programs).  WITH THE NEW BUILT-IN FILTERS, the decoder becomes more easily consumer-usable on Windows.  (Easier for me most of the time also.)

 

MORE GOOD NEWS -- with the great new filter sequence structure, it is much simplier to 'zero in' on proper decoding in the less common cases.  I have the base filters and calibration numbers now for 'Sergio Mendes/Brasil'66' and 'Herb Alpert & Tijuana Brass' albums!!!   The tonality of the instruments and much more natural rendition of the vocals is very nice.   I intend full decodes on the review group and public snippets.   These are NOT the 'breakthroughs' like Soul Kiss or SuperTramp decodes, but are very nice indeed.

 

It takes several hours to do the decodes, and I am hoping to get something posted before the end of the workday today.  (I have some family cooking to do -- very distracting when working with raw chicken as cleanliness is paramount and I share the kitchen with an older mother who cannot withstand infections/bacteria... My cooking quality standards are laboratory clean and very tedious.  Gotta be very careful!!!  I make the most beautiful and satisfying chicken soup... )

 

John

 

 

 

[Don Hills]

John,

Your filter sets do point to a way forward. With the decoder and a list of settings, others can process their own copies of the tracks you have analysed so far. Armed with these for before/after comparison, those who are interested can learn how to recognise the processing (mis)applied to other tracks and decode them. Settings can be freely shared, unlike the processed tracks themselves.

[fas42]

16 hours ago, John Dyson said:

You spend 10 minutes at most thinking about the problem, I spend 4yrs studying/solving the problem...  Think about the likelihood of who is being accurate or not.  Spend some time really trying to understand the problem -- then we can talk...

 

John, the problem as you see it is that numerous recordings are poorly mastered - and you have worked out a technical solution, that make those recordings sound better, to you. I have zero problems with your expertise in achieving that, and salute your efforts.

 

The problem as I see it is that nearly all audio systems suffer enough audible problems to make many recordings unpleasant to listen to - and I have been aware for 35 years of what's going on, and looking at various aspects of this. If there were enough setups out there that were good enough to point to a better future I wouldn't be so vocal about it; but there aren't - so while people keep rubbishing recordings that are perfectly satisfactory if replayed on a good enough rig, I'll keep responding.

 

You think I'm wrong; I think that you have worked out one strategy to compensate for the general lack of sufficient capability of current playback systems - how about we leave it at that?

[sandyk]

33 minutes ago, Don Hills said:

John,

Your filter sets do point to a way forward. With the decoder and a list of settings, others can process their own copies of the tracks you have analysed so far. Armed with these for before/after comparison, those who are interested can learn how to recognise the processing (mis)applied to other tracks and decode them. Settings can be freely shared, unlike the processed tracks themselves.

 Don

 This sounds like a good idea, however John has developed his Decoder in collaboration with Richard, and there then become commercial considerations involved. It may however be possible further down the track to purchase a licensed copy from Richard.

 With somebody who has a vast collection of CDs from that period it could be a very worthwhile solution.

 

Alex

[sandyk]

4 minutes ago, fas42 said:

You think I'm wrong; I think that you have worked out one strategy to compensate for the general lack of sufficient capability of current playback systems - how about we leave it at that?

 Frank

 Please do so .

You have contributed very little worthwhile to the discussion as you do not currently have a high enough quality system

available to properly evaluate the results from the decoder, unless you do something like saving the comparisons to USB memory and take the USB stick to your audiophile friend's house. Both of you could then sit down and listen to before and after versions via his system, and you then take on board his comments as well..

[John Dyson]

30 minutes ago, sandyk said:

 Don

 This sounds like a good idea, however John has developed his Decoder in collaboration with Richard, and there then become commercial considerations involved. It may however be possible further down the track to purchase a licensed copy from Richard.

 With somebody who has a vast collection of CDs from that period it could be a very worthwhile solution.

 

Alex

Don's idea is good...   I agree with you also Alex about considering Richards' position and opinon (of course.)

This is a rambling answer -- I have been thinking about these issues -- I want to fix the music -- here are my thoughts:

------

 

In my discussions with Richard (and I don't think I am divulging anything that he would feel uncomfortable), he is skeptical of a consumer product for several reasons.  Perhaps the most practical problem  is that of customer support.  Even decoding a master tape with tones, etc -- using the DHNRDS DA isn't a fun thing.*

 

* On Linux, with the correct scripts, the DHNRDS is trivial to use, but most people aren't using Linux and instead live in the GUI world.  Currnetly, the DHNRDS DA is not GUI.  Because of my deep Unix roots, I wrote the DHNRDS DA as command line, even though a GUI interface has been discussed.  I am missing the brain cells to do GUI code (really)...  I am graphically challenged.

 

When a product costs maybe $300 in a commercial setting, it is dirt cheap and just a nuisance fee.  In a consumer setting, it is expensive and demands support.  Consumers will tend to demand more support than a professional who has generally had some experience using a real DolbyA unit and is used to running albums when needed.  Kindly online support like though email or the forums works for very small numbers of users, but for a lot of users, with the very archane user interface, I can forsee a major train wreck -- just as Richard might predict.

 

I have tried to walk the middle ground for now -- until there is a better solution, where I offer 'trial' versions to the consumer, using the same license as commercial trial versions where I create license files with 2-3 month timeouts.   In both cases (consumer and commercial) the trial version is free to use up until the timeout.  For consumers, I have planned to offer new updated license timeouts ad-infinitum.  Since volume is so very small right now, I can keep up with license requests.

 

After at least I get done with the C4, I might add the filter mechanism to the DA decoder.  Currently, I use filter sequences built using SOX commands inside of shell scripts.  Those filter sequencies are approximate, but generally close enough to avoid obvious artifacts.

 

A 'temporary' license for those with big CD collections JUST MIGHT be useful -- esp with the built-in 'filter packs'.   Even better, the built-in filter packs can be more accurate than the filter sequences that I manually build.  In a way, I am such a user with a moderately large digital collection and work towards a complete decoded collection.  Once I have all of the usage kinks worked out, maybe I can put together a nice, coherent package that hides the nasty parts, and simiply does the consumer decode...  Except for choosing the correct 'filter pack' and the 'calibration level', the decoder could be a black box.   BTW -- since the decoder is now rock solid, I have found that most albums are encoded with very similar 'calibration levels' -- so the bugaboo is the 'filter pack'.  Secondarily, using command lines on native Windows is not a 'fun' experience to say the least.   Richard and I have discussed a GUI scheme that avoids GPL licensing issues, also isolates the GUI code that is so difficult for me.  If we get enough consumer interest maybe we can put together a consumer decoder version.

 

When trying to sell a consumer/commercial product, there is also the matter of product differentiation.  I have thought about a 'free' consumer version that decodes better than any DolbyA, but not as good as the 'commercial' decoder version.   There isn't that big of a difference when using the modulation distortion mitigation version of gain control vs. simple multiplication, but there is enough that pros would like the smoother sound of the full decoder.  A benefit of running with the anti-MD code disabled -- the decoder can be made to run much faster.  If someone has a lot of successful experience with the consumer decoder, then it becomes much less risky to actually sell the commercial version to them.  This kind of behavior seems that it might be illegal or just bad karma -- not sure if it is workable..

 

My REAL/ideal goal is to kick the distributors in the butt -- so that they sell properly mastered  stuff.  My product volumes will always be vanishingly small -- I simple wish to create a legacy of improving the listening experience.   If the sound quality improvement is implemented the level of distribution or at the level of the end user -- I would still be very happy!!!

 

As always, I give time-out versions gratis...  I would be happier to give them away when the filter pack scheme is created as I would be more comfortable that users would be more satisfied customers.

 

John

[fas42]

1 hour ago, sandyk said:

 Frank

 Please do so .

You have contributed very little worthwhile to the discussion as you do not currently have a high enough quality system

available to properly evaluate the results from the decoder, unless you do something like saving the comparisons to USB memory and take the USB stick to your audiophile friend's house. Both of you could then sit down and listen to before and after versions via his system, and you then take on board his comments as well..

 

Alex, you still don't don't get it .. so-called "high enough quality system"s normally exaggerate distortion anomalies, and make so many recordings unpleasant to listen to - one solution is to tame those aspects of a recording that provoke a system to audibly misbehave, as described in this thread.

 

What we do down the road is work on eliminating remaining issues with his rigs, so that completely conventional recordings deliver good sound - a bargain basement CD of ELO came up trumps last round - I was impressed!

 

Before and after versions are merely different masterings - and his SQ is good enough to make audiophile reissues come across as boring, watered down versions of the originals.

[John Dyson]

10 minutes ago, fas42 said:

 

Alex, you still don't don't get it .. so-called "high enough quality system"s normally exaggerate distortion anomalies, and make so many recordings unpleasant to listen to - one solution is to tame those aspects of a recording that provoke a system to audibly misbehave, as described in this thread.

 

What we do down the road is work on eliminating remaining issues with his rigs, so that completely conventional recordings deliver good sound - a bargain basement CD of ELO came up trumps last round - I was impressed!

 

Before and after versions are merely different masterings - and his SQ is good enough to make audiophile reissues come across as boring, watered down versions of the originals.

The DHNRDS DA provides the closest to what was recorded by the artist/original recording engineer as possible today.   The junk being sold today with all kinds of effectve 'tweaked tone control', not very close to what was recorded at the studio.  Fact -- not opinon.

 

I don't want anyone to tweak my tone control -- I can do it myself, and most people will tweak the tone control rather than some unintended serendipity being distributed on CDs today.

 

John

 

[sandyk]

12 minutes ago, fas42 said:

Alex, you still don't don't get it

 

No Frank, you don't get it. Your experiences are completely at odds with the vast majority of members.

 Do what I suggested about saving to USB then trying John's before and after versions in a system far better than the one that you are currently using.

[John Dyson]

BTW -- Alex, I have a big backlog of good decodes now.  I am doing Carly Simon's colleciton again -- noticeable further improvement.  The 'Crime of the Century' and 'Breakfast in America' are done.  I am still not happy with Linda Ronstadts' pop hits, but the wonderful band stuff is amazing.  Let me know if you hear signifcant defects.  I think that there are a few possible choices left, I tried to make the best ones that I could discern.  I might go ahead and process 'In the Quietest Moments'.  Oh -- also I have GOOD Brasil'66 and Herb Alpert decodes also -- the instruments sound really good.  The Brasil'66 vocals are still a little 'enhanced' somehow., but better than ever.  Listening to the regular 'Herb Alpert' release -- it is hissy and compressed -- might even be compressed DolbyA -- YUCK!!!
Still have the totally *perfect* ABBA and Carpenters decodes (all of the actual albums) complete also.

 

 I'll be setting up reviews on the review group later on -- probably tomorrow.  Some serious QC is in order.

 

Take care!!!  Will probably discuss more privately later on.  Let me know anything you might be interested -- this goes also for people in the review group.

 

John

 

[sandyk]

3 minutes ago, John Dyson said:

BTW -- Alex, I have a big backlog of good decodes now.  I am doing Carly Simon's colleciton again -- noticeable further improvement.  The 'Crime of the Century' and 'Breakfast in America' are done.  I am still not happy with Linda Ronstadts' pop hits, but the wonderful band stuff is amazing.  Let me know if you hear signifcant defects.  I think that there are a few possible choices left, I tried to make the best ones that I could discern.  I might go ahead and process 'In the Quietest Moments'.  Oh -- also I have GOOD Brasil'66 and Herb Alpert decodes also -- the instruments sound really good.  The Brasil'66 vocals are still a little 'enhanced' somehow., but better than ever.  Listening to the regular 'Herb Alpert' release -- it is hissy and compressed -- might even be compressed DolbyA -- YUCK!!!
Still have the totally *perfect* ABBA and Carpenters decodes (all of the actual albums) complete also.

 

Take care!!!  Will probably discuss more privately later on.

 

John

 

Hi John

 I presume that these will also be made available to other participants in your current PM series ?

 Anybody else interested in helping with constructive comments could also contact you directly via PM about being included in this group.

 

Kind Regards

Alex

[John Dyson]

Just now, sandyk said:

Hi John

 I presume that these will also be made available to other participants in your current PM series ?

 Anybody else interested in helping with constructive comments could also contact you directly via PM about being included in this group.

 

Kind Regards

Alex

Yes -- I am planning to put some together for review tomorrow/friday at the latest -- I am not excluding anyone in the review group at all.  People interested to review should let me know.  I'll also do the same on your forum.

 

Of course, the QC on my side takes time, and I like to seriously review after a day or so -- rather than burning my hearing out thinking that everything sounds good...  I have made that mistake too often.

 

I am esp hopeful about the Supertramp and Carly stuff (other than the decoding that I did for you personally -- it was really special.)  Some of the original Carpenters stuff was special, of course there are others also.  Actually, the originally nasty sounding Carpenters download from HDtracks, after decoding, sounds really good!!!  I'll throw in a few of those songs just as a bonus -- they are a REAL bonus.

 

John

 

[phosphorein]

12 minutes ago, John Dyson said:

BTW -- Alex, I have a big backlog of good decodes now.  I am doing Carly Simon's colleciton again -- noticeable further improvement.  The 'Crime of the Century' and 'Breakfast in America' are done.  I am still not happy with Linda Ronstadts' pop hits, but the wonderful band stuff is amazing.  Let me know if you hear signifcant defects.  I think that there are a few possible choices left, I tried to make the best ones that I could discern.  I might go ahead and process 'In the Quietest Moments'.  Oh -- also I have GOOD Brasil'66 and Herb Alpert decodes also -- the instruments sound really good.  The Brasil'66 vocals are still a little 'enhanced' somehow., but better than ever.  Listening to the regular 'Herb Alpert' release -- it is hissy and compressed -- might even be compressed DolbyA -- YUCK!!!
Still have the totally *perfect* ABBA and Carpenters decodes (all of the actual albums) complete also.

 

 I'll be setting up reviews on the review group later on -- probably tomorrow.  Some serious QC is in order.

 

Take care!!!  Will probably discuss more privately later on.  Let me know anything you might be interested -- this goes also for people in the review group.

 

John

 

BTW John, I had a listen to “The Logical Song;” what an eye-opener! You’ve done a marvelous job with your decoder.

 

[Samuel T Cogley]

7 minutes ago, phosphorein said:

BTW John, I had a listen to “The Logical Song;” what an eye-opener! You’ve done a marvelous job with your decoder.

 

 

Based on this post, I fetched the demo files.  I guess I didn't realize that "proper" decoding seems to restore some dynamic range (it certainly looks that way on the waveform). This is a superb example!

[sandyk]

1 hour ago, John Dyson said:

I am esp hopeful about the Supertramp and Carly stuff (other than the decoding that I did for you personally -- it was really special.) 

 It was indeed, and not only was it way better sounding than the original CD, it then was also markedly superior to the official 24/192 version despite being from a 16/44.1 source. 

PM me for a couple of samples (You're So Vain, and We Have No Secrets) They are .flac in a 24/88.2 container

[John Dyson]

6 hours ago, sandyk said:

 It was indeed, and not only was it way better sounding than the original CD, it then was also markedly superior to the official 24/192 version despite being from a 16/44.1 source. 

PM me for a couple of samples (You're So Vain, and We Have No Secrets) They are .flac in a 24/88.2 container

 

ADD-ON:  the comments below refer to the NEW decodes...  The decodes that Alex has are pretty good -- I am doing further tweaking to mitigate very minor issues...  I'll be making the new decodes available as soon as they pass muster...

 

There are two major tricks to getting the decode EQ correct, and one additional repeated mistake that I make.

 

My repeated mistake is that I tend to add more bass at 750 and below than should be.  Some material needs it to balance out the LF distortion, but most material doesn't.  DolbyA produces significant LF distortion in the 20-150Hz range unless properly decoded.  Once the DolbyA is decoded, then the distortion is cancelled -- but if the LF EQ is too far off, then there will be residual or even worse LF distortion!!!

 

The actial 'tricks' are related to what needs to happen in the MF frequency range (between 1kHz and 3kHz), and where to place the HF knee.   The corrective EQ usually requires a dip in the 1k to 3kHz range, a relative peak at the 3kHz range, and then to fall off beyond 3 or 4.5kHz.

 

Corrrecting the depth of the MF dip and how quickly to roll off the HF -- those are the sensitive adjustments.  Note that there is no real loss of HF, because there is a relative peak at usually 3kHz or the 3kHz->4.5kHz range, and then a fall-off.  Usually, the eventual value of the fall-off in the 9kHz range equals the gain boost in the 3kHz range, so all is pretty much compensated.

 

The new 'Carly' has better MF compensation and a more carefully adjusted calibration level.  It is all so very trivial once the concepts are understood and the general ranges of EQ are known!!!

 

Just restarted Carly -- did my usual thing, mis-transcribed the EQ -- ended up with something that wasn't quite right.  Restarted the decode again.  Also, similar mistake on 'Crime'.  This is why I try to do a 'quality control' check a few days after the actual decode operation...

 

John

 

[Samuel T Cogley]

Apologies if this has been discussed and I missed it.

 

With all the boutique, "audiophile" remasters that have been produced since the '90s on labels such as DCC, MFSL, and Audio Fidelity (there are others, these are ones that I personally own and am most familiar with), I'm now really starting to wonder if we were "sold a bill of goods" with regard to the alleged White Glove treatment this remasters received.  If the Dolby A decoding was not applied (and it seems in many cases it was NOT), can we really say the remaster was even competent, let alone, "audiophile quality"?

 

Perhaps @John Dyson can either elaborate on his understanding of Dolby A decoding (or lack of) in those "boutique" settings, or direct me to a post/thread where it's discussed.

 

EDIT: Additionally, since Dolby is an ongoing concern, don't they bear the bulk of the culpability for the lack of "correct" Dolby A decoding in the field?

 

Thanks

 

 

[John Dyson]

1 hour ago, Samuel T Cogley said:

Apologies if this has been discussed and I missed it.

 

With all the boutique, "audiophile" remasters that have been produced since the '90s on labels such as DCC, MFSL, and Audio Fidelity (there are others, these are ones that I personally own and am most familiar with), I'm now really starting to wonder if we were "sold a bill of goods" with regard to the alleged White Glove treatment this remasters received.  If the Dolby A decoding was not applied (and it seems in many cases it was NOT), can we really say the remaster was even competent, let alone, "audiophile quality"?

 

Perhaps @John Dyson can either elaborate on his understanding of Dolby A decoding (or lack of) in those "boutique" settings, or direct me to a post/thread where it's discussed.

 

EDIT: Additionally, since Dolby is an ongoing concern, don't they bear the bulk of the culpability for the lack of "correct" Dolby A decoding in the field?

 

Thanks

 

 

Regarding proper decoding on labels like MFSL...  It seems to be hit or miss.  There are some technical hints that help to complete the evaluation -- one fairly strong indicator is a larger than typical noise background on a spectogram, and often a slight increase in noise vs freq somwhere above 10kHz.  If there is more hiss than expected on commercially released material, the probability of the material being DolbyA is pretty high, it is visible even after the EQ used for mastering.   Very very old material WILL have a lot of hiss even if not DolbyA., but other aspects of the Nat King Cole remasters APPEAR to have the DolbyA character.  I'll try a decode of Nat again in the next day -- I am curious to prove/disprove myself...

 

Another data point, I have an MFSL remaster of a Carly Simon album, and it doesn't sound right.  Not 100% sure if it is DolbyA or not, but will check later on with some detail.  There IS something wrong with the mastering though.  (it is a 2015 remaster of the Carly Simon album, the one with 'The Way I'v Always Heard It Should Be')   Later, I could take a look at the metadata/album cover to give you more info if it would be helpful.   The copy 'Heard it should be' recently produced in the DHNRDS DA decode seems to be more clean, less grainy.   I think that Alex can give you feedback on that one.

 

AFZ5-197  Bread album is most definitely DolbyA without decoding.  As I wrote above -- hit or miss.  Previously, I demoed the Bread stuff -- but my decodes were faulty beacause of using the wrong filter formulas -- I didn't know the correct structure/template yet.   My most recent experimental decodes show with 100% probability that the Bread stuff is left in encoded state.  I'll add some Bread songs to the demo snippets.  The Bread album does sound better when *properly* decoded, but the undecoded compressed sound isn't really all that bad .  (The EQ instead of decoding is 'tolerable' in the case of the Bread album.)

 

I also have a Supertramp remaster that is perfect -- decoded properly and everything (I can give you an ID when I have a chance -- I just added this note, and fixing spaghetti right now -- took a quick break :-)).

 

Sources like HDtracks are even less of a guarantee -- I have both the unlmited version of Band On the Run by McCartney, and it is DolbyA without decoding, and a BEAUTIFUl-AFTER-DECODING copy of the remastered singles for the Carpenters... The Carpenters snippet examples are in the demo set.

 

-----

About the culpability -- responsibility is NOT with the Dolby company, the cupability is with the distributors of the recordings.  The tapes (or even digital copies) of the recordings are often kept in a DolbyA encoded state.  When producing the vinyl or CD, there should be a DolbyA decoding step.  Decoding the material is a choice of the distributor.  I cannot even blame the remastering engineers like Hoffman, because the traditional process would be to deliver the tape (or digital copy) in DolbyA form. It is the distributor who doesn't prepare the tape (or recording) for distribution.

 

The Library of Congress proceedures for archiving old recordings does NOT mandate NR decoding the material.  It is an *optional* step.  The intent is that the decoding should be done immediately before further processing or creationing the version for delivery to the customer.  This means that a lot of digital archives are in DolbyA form.  Frustrating to the production process -- the DolbyA HW only runs in realtime, much slower than a digital file copy.  The DHNRDS DA could solve that problem if ever added to the process,  but I am definitley not expecting that to happen.

 

---------------------------------

About DolbyA encoding/decoding

 

Even though the DolbyA HW encoding is nearly perfect, the decoding process with the old DolbyA HW (A301, cat22 or whatever) is only approximate.   By using the DHNRDS DA, the resulting sound quality is noticeably better.  (The DHNRDS DA is not well tested in multi-generation encodes/decodes, but for a single decode it blows away a real DolbyA.)  The DHNRDS DA doesn't depend on an approximate inverse calculation (delays that aren't important in a forward signal flow are much more important when that flow is in a feedback loop -- that delay in a feedback loop is one reason why the DolbyA HW cannot be as accurate as the DHNRDS DA.)

So, this sad thing about selling DolbyA encoded material through the years does have a silver lining.  When the old recordings are finally decoded -- even by the consumer like me -- with a DHNRDS DA, the sound quality can be better that would have otherwise been!!!   The bad thing is that industry doesnt care and it appears that most consumers don't care either...  (Sorry for the editorial, but I am trying to be honest in the explanation about what is going on.)

 

* I will give definititive answers on my the premium recordings in my collection JUST TO DOUBLE CHECK.  My previous evaluations are probably true, but I never fully trust myself...

 

John

 

 

 

 

[Samuel T Cogley]

51 minutes ago, John Dyson said:

About the culpability -- responsibility is NOT with the Dolby company, the cupability is with the distributors of the recordings.  The tapes (or even digital copies) of the recordings are often kept in a DolbyA encoded state.  When producing the vinyl or CD, there should be a DolbyA decoding step.  Decoding the material is a choice of the distributor.  I cannot even blame the remastering engineers like Hoffman, because the traditional process would be to deliver the tape (or digital copy) in DolbyA form. It is the distributor who doesn't prepare the tape (or recording) for distribution.

 

The Library of Congress proceedures for archiving old recordings does NOT mandate NR decoding the material.  It is an *optional* step.  The intent is that the decoding should be done immediately before further processing or creationing the version for delivery to the customer.  This means that a lot of digital archives are in DolbyA form.  Frustrating to the production process -- the DolbyA HW only runs in realtime, much slower than a digital file copy.  The DHNRDS DA could solve that problem if ever added to the process,  but I am definitley not expecting that to happen.

 

Thank you John for all that!!!

 

For some reason, I was under the impression that lack of Dolby A decoding was due to lack of hardware decoder availability.  I was thinking that the boutique houses didn't want to purchase the hardware shown at the 2:00 mark on this video (just an example).

 

EDIT: If the hardware isn't available anymore, isn't that Dolby's fault?

 

Using that video as an example workflow, the decoding is part of the digital capture process (but in the analog domain).  I think I hear you saying that capturing un-decoded and then applying the decoding in the digital domain produces better results.  Does Dolby make such a digital decoder?  I'm thinking the answer is no, based on all your work here.

[John Dyson]

57 minutes ago, Samuel T Cogley said:

 

Thank you John for all that!!!

 

For some reason, I was under the impression that lack of Dolby A decoding was due to lack of hardware decoder availability.  I was thinking that the boutique houses didn't want to purchase the hardware shown at the 2:00 mark on this video (just an example).

 

EDIT: If the hardware isn't available anymore, isn't that Dolby's fault?

 

Using that video as an example workflow, the decoding is part of the digital capture process (but in the analog domain).  I think I hear you saying that capturing un-decoded and then applying the decoding in the digital domain produces better results.  Does Dolby make such a digital decoder?  I'm thinking the answer is no, based on all your work here.

Supposedly, there is still significant DolbyA HW available on places like EBAY.  DolbyA units don't fail unless modified, unlike Telcom C4 units which drift away (reason for pressure for me to write a C4 decoder.)  (There had been a tradition to modify DolbyA units for vocal enhancement -- e.g. brightening Karen Carpenter's vocals.)

 

Dolby did decide not to make any more -- they told us that we wouldn't be competing with them.  They have made a cinema DolbyA that is supposedly done in software, but cinema has simpler requirements than true high fidelity decoding.  It would be VERY surprising if they had a working DolbyA decoder in SW 20yrs ago that sounded very good by today's audiophile standards!!!   Also, there is a 'plug-in' 3rd party decoder (forget the name), but it doesn't even sound like a DolbyA, and the NR isn't very effective.  My project partner rejected the plug-in way way way before he even met me or considered being involved in creating a set of NR systems.

 

The BIG problem with normally available real Dolby HW is that it isn't digital friendly, and requires analog I/O.  Also, requiring analog I/O and living in the analog world could work in realtime only.

 

The BIG problem with the DHNRDS is that it isn't analog tape recorder friendly -- it doesn't integrate with the professional analog tape equipment, while the true DolbyA does so very nicely.

 

The DHNRDS style decoder can run faster than relatime and can be set-up easily for batch decodes.

 

Even though I wrote the DHNRDS to have the highest sound quality possible, the actual goal of the DHNRDS project was simply to work as well as DolbyA HW.  In a way, I missed the target a bit, because the DHNRDS could have run MUCH faster using the algorithms that I had developed (very non-obvious) if I hadn't added all of the anti-modulation distortion capabilities.

 

John

 

ADD-ON:  I do decodes for people gratis.  I happily destroy source material and resulting decodes if desired or required by law.  It would be SO VERY NICE if the customer base would be able to start pressuring for better quality.  Makes no real difference (other than minor quality issues) to use DHNRDS or DolbyA, but the customers should really be getting properly mastered material (just IMO.)

 

[John Dyson]

Happy time -- got some new decodes done...   My Carpenters DolbyA copies have 'popped'.  I figured out they were doing some crazy stuff to the files that people normally listen to.  I'll go into the details, but it is CRAZY...  One thing that they did was to DolbyA encode in Mid/Side mode, yet use the files in Left/Right mode.  (There is a method to the Mid/Side madness -- it has to do with the levels of 'Side' are low enough that the DolbyA isn't very active -- less DolbyA distortion.)   I have only done 1971 (Carpenters) and 1972 (A Song For You) in the maximum quality modes.   Note that the quality of 1969, 1970, 1971 and less of a problem 1972 aren't all that great.  1973 is beautiful, but I am not done yet...

https://www.dropbox.com/sh/jt0er1glod12ego/AAC3qMSgSKH3Kkc-g8_imGpEa?dl=0

 

(These are necessarily snippets -- full copies available in the review group.  ABBA has 'popped' also -- upcoming.)

 

[John Dyson]

Sorry -- had some problems -- all fixed, and very nice...  Just got 1971 done onto the demo site, and now should be uploading 1972 in a few minutes, and 1973 in about 1/2 Hr.

 

[sandyk]

3 hours ago, John Dyson said:

Sorry -- had some problems -- all fixed, and very nice...  Just got 1971 done onto the demo site, and now should be uploading 1972 in a few minutes, and 1973 in about 1/2 Hr.

 

 

Very nice doesn't even come close to describing John's just completed test of a few corrected tracks from Supertramp.

 They absolutely put to shame most Hi Res recordings despite being sourced from 16/44.1 .

[John Dyson]

9 hours ago, sandyk said:

 

Very nice doesn't even come close to describing John's just completed test of a few corrected tracks from Supertramp.

 They absolutely put to shame most Hi Res recordings despite being sourced from 16/44.1 .

 

I want to make *very clear* that some recordings 'can' really benefit from better processing because they are fairly good to begin with.  Also, there are some recordings which will never be perfect, but commonly available copies are still lacking in quality.  For example, the Carpenters recordings will NEVER--EVER  present as 2000's perfection.  In fact, I hear defects in even some of the better recordings...  A lot of defects are somes on the original master tape!!!

 

Where Alex and I might disagree -- the Supertramp Crime of the Century is an example of really really really good '70s technology, and the music is 'unique' (at least to me.)  I had (still have) all kinds of troubles knowing what 'Crime of the Century' should sound like.   When I listen to it, I keep on wanting to use a 3dB treble cut at 9kHz with a Q=0.8409.  Why do I want to get rid of some of the highs?  The vocals have an edge that sounds like a sharp treble boost to me.  That treble boost (or seems so) triggered a problem in the DA decoder, which was REALLY a problem.  It made those hard edges in the vocals cause the DA decoder gain to bounce -- thereby worsening the edges.  The fact that those hard edges caused a problem was an ACTUAL DHNRDS DA bug.  (In fact, there was also an attack/release time bug where it was about 6% too fast -- thereby making ambience disappear in very limited cases.)  However, that 'bug' isn't where I was having bad problems getting a good decode.  Without Alex nudging me on, I might not have fixed the DHNRDS bug or have been able to produce something that an audiophile might expect.

 

Not knowing how material is REALLY supposed to sound makes the 'very high quality' decoding very very difficult to do.  This is where disagreeing amongst friends can really help.  Frankly, I didn't know what to do about the problem.  We 'compromised' by myself deciding that 'the customer is right' -- which means, I am going to trust the 'audiophiles' opinion, because it is really wrong to be driven by technical reasons for a certain sound.

 

My eventual compromise is that I do admit that the current version (that Alex rightfully likes) is the official sound, and does sound similar to some of the better commercial examples.   For my own listening, I do at least 1 filter: -3dB@9k/Q=0.8409.  Then, we are all happy :-).

 

This stuff IS very difficult, and it requires that I get help from people not just willing, but interested in participating.  I cannot do it all myself, because I don't have all of the needed skills.  It is really nice that Alex is willing to help/participate, and as a side benefit, in about an hour or so, an entire copy of the album ready for final evaluation,  will be available to Alex and other testers.

 

John

[sandyk]

 I should also point out, that in a couple of cases where I have felt that the latest version has lost something when trying to correct for excessive sibilance that may not have been as bad on the original Master  recording, I ask a couple of friends of mine from overseas, one of whom is a performing classical musician from the USA who is resident in Germany, to check my findings, the results of which are then directly forwarded to John. In the case of Supertramp, it is also helpful in  that both of these friends have the Supertramp album on Vinyl for comparison.